0001: #ifndef _LINUX_SCHED_H
0002: #define _LINUX_SCHED_H
0003:
0004: /*
0005: * cloning flags:
0006: */
0007: #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
0008: #define CLONE_VM 0x00000100 /* set if VM shared between processes */
0009: #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
0010: #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
0011: #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
0012: #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
0013: #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
0014: #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
0015: #define CLONE_THREAD 0x00010000 /* Same thread group? */
0016: #define CLONE_NEWNS 0x00020000 /* New namespace group? */
0017: #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
0018: #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
0019: #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
0020: #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
0021: #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
0022: #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
0023: #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
0024: /* 0x02000000 was previously the unused CLONE_STOPPED (Start in stopped state)
0025: and is now available for re-use. */
0026: #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
0027: #define CLONE_NEWIPC 0x08000000 /* New ipcs */
0028: #define CLONE_NEWUSER 0x10000000 /* New user namespace */
0029: #define CLONE_NEWPID 0x20000000 /* New pid namespace */
0030: #define CLONE_NEWNET 0x40000000 /* New network namespace */
0031: #define CLONE_IO 0x80000000 /* Clone io context */
0032:
0033: /*
0034: * Scheduling policies
0035: */
0036: #define SCHED_NORMAL 0
0037: #define SCHED_FIFO 1
0038: #define SCHED_RR 2
0039: #define SCHED_BATCH 3
0040: /* SCHED_ISO: reserved but not implemented yet */
0041: #define SCHED_IDLE 5
0042: /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
0043: #define SCHED_RESET_ON_FORK 0x40000000
0044:
0045: #ifdef __KERNEL__
0046:
0047: struct sched_param {
0048: int sched_priority;
0049: };
0050:
0051: #include <asm/param.h> /* for HZ */
0052:
0053: #include <linux/capability.h>
0054: #include <linux/threads.h>
0055: #include <linux/kernel.h>
0056: #include <linux/types.h>
0057: #include <linux/timex.h>
0058: #include <linux/jiffies.h>
0059: #include <linux/rbtree.h>
0060: #include <linux/thread_info.h>
0061: #include <linux/cpumask.h>
0062: #include <linux/errno.h>
0063: #include <linux/nodemask.h>
0064: #include <linux/mm_types.h>
0065:
0066: #include <asm/system.h>
0067: #include <asm/page.h>
0068: #include <asm/ptrace.h>
0069: #include <asm/cputime.h>
0070:
0071: #include <linux/smp.h>
0072: #include <linux/sem.h>
0073: #include <linux/signal.h>
0074: #include <linux/compiler.h>
0075: #include <linux/completion.h>
0076: #include <linux/pid.h>
0077: #include <linux/percpu.h>
0078: #include <linux/topology.h>
0079: #include <linux/proportions.h>
0080: #include <linux/seccomp.h>
0081: #include <linux/rcupdate.h>
0082: #include <linux/rculist.h>
0083: #include <linux/rtmutex.h>
0084:
0085: #include <linux/time.h>
0086: #include <linux/param.h>
0087: #include <linux/resource.h>
0088: #include <linux/timer.h>
0089: #include <linux/hrtimer.h>
0090: #include <linux/task_io_accounting.h>
0091: #include <linux/latencytop.h>
0092: #include <linux/cred.h>
0093: #include <linux/llist.h>
0094:
0095: #include <asm/processor.h>
0096:
0097: struct exec_domain;
0098: struct futex_pi_state;
0099: struct robust_list_head;
0100: struct bio_list;
0101: struct fs_struct;
0102: struct perf_event_context;
0103: struct blk_plug;
0104:
0105: extern int disable_nx;
0106: extern int print_fatal_signals;
0107:
0108: /*
0109: * List of flags we want to share for kernel threads,
0110: * if only because they are not used by them anyway.
0111: */
0112: #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
0113:
0114: /*
0115: * These are the constant used to fake the fixed-point load-average
0116: * counting. Some notes:
0117: * - 11 bit fractions expand to 22 bits by the multiplies: this gives
0118: * a load-average precision of 10 bits integer + 11 bits fractional
0119: * - if you want to count load-averages more often, you need more
0120: * precision, or rounding will get you. With 2-second counting freq,
0121: * the EXP_n values would be 1981, 2034 and 2043 if still using only
0122: * 11 bit fractions.
0123: */
0124: extern unsigned long avenrun[]; /* Load averages */
0125: extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
0126:
0127: #define FSHIFT 11 /* nr of bits of precision */
0128: #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
0129: #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
0130: #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
0131: #define EXP_5 2014 /* 1/exp(5sec/5min) */
0132: #define EXP_15 2037 /* 1/exp(5sec/15min) */
0133:
0134: #define CALC_LOAD(load,exp,n) \
0135: load *= exp; \
0136: load += n*(FIXED_1-exp); \
0137: load >>= FSHIFT;
0138:
0139: extern unsigned long total_forks;
0140: extern int nr_threads;
0141: DECLARE_PER_CPU(unsigned long, process_counts);
0142: extern int nr_processes(void);
0143: extern unsigned long nr_running(void);
0144: extern unsigned long nr_uninterruptible(void);
0145: extern unsigned long nr_iowait(void);
0146: extern unsigned long nr_iowait_cpu(int cpu);
0147: extern unsigned long this_cpu_load(void);
0148:
0149:
0150: extern void calc_global_load(unsigned long ticks);
0151: extern void update_cpu_load_nohz(void);
0152:
0153: extern unsigned long get_parent_ip(unsigned long addr);
0154:
0155: struct seq_file;
0156: struct cfs_rq;
0157: struct task_group;
0158: #ifdef CONFIG_SCHED_DEBUG
0159: extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
0160: extern void proc_sched_set_task(struct task_struct *p);
0161: extern void
0162: print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
0163: #else
0164: static inline void
0165: proc_sched_show_task(struct task_struct *p, struct seq_file *m)
0166: {
0167: }
0168: static inline void proc_sched_set_task(struct task_struct *p)
0169: {
0170: }
0171: static inline void
0172: print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
0173: {
0174: }
0175: #endif
0176:
0177: /*
0178: * Task state bitmask. NOTE! These bits are also
0179: * encoded in fs/proc/array.c: get_task_state().
0180: *
0181: * We have two separate sets of flags: task->state
0182: * is about runnability, while task->exit_state are
0183: * about the task exiting. Confusing, but this way
0184: * modifying one set can't modify the other one by
0185: * mistake.
0186: */
0187: #define TASK_RUNNING 0
0188: #define TASK_INTERRUPTIBLE 1
0189: #define TASK_UNINTERRUPTIBLE 2
0190: #define __TASK_STOPPED 4
0191: #define __TASK_TRACED 8
0192: /* in tsk->exit_state */
0193: #define EXIT_ZOMBIE 16
0194: #define EXIT_DEAD 32
0195: /* in tsk->state again */
0196: #define TASK_DEAD 64
0197: #define TASK_WAKEKILL 128
0198: #define TASK_WAKING 256
0199: #define TASK_STATE_MAX 512
0200:
0201: #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
0202:
0203: extern char ___assert_task_state[1 - 2*!!(
0204: sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
0205:
0206: /* Convenience macros for the sake of set_task_state */
0207: #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
0208: #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
0209: #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
0210:
0211: /* Convenience macros for the sake of wake_up */
0212: #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
0213: #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
0214:
0215: /* get_task_state() */
0216: #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
0217: TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
0218: __TASK_TRACED)
0219:
0220: #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
0221: #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
0222: #define task_is_dead(task) ((task)->exit_state != 0)
0223: #define task_is_stopped_or_traced(task) \
0224: ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
0225: #define task_contributes_to_load(task) \
0226: ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
0227: (task->flags & PF_FREEZING) == 0)
0228:
0229: #define __set_task_state(tsk, state_value) \
0230: do { (tsk)->state = (state_value); } while (0)
0231: #define set_task_state(tsk, state_value) \
0232: set_mb((tsk)->state, (state_value))
0233:
0234: /*
0235: * set_current_state() includes a barrier so that the write of current->state
0236: * is correctly serialised wrt the caller's subsequent test of whether to
0237: * actually sleep:
0238: *
0239: * set_current_state(TASK_UNINTERRUPTIBLE);
0240: * if (do_i_need_to_sleep())
0241: * schedule();
0242: *
0243: * If the caller does not need such serialisation then use __set_current_state()
0244: */
0245: #define __set_current_state(state_value) \
0246: do { current->state = (state_value); } while (0)
0247: #define set_current_state(state_value) \
0248: set_mb(current->state, (state_value))
0249:
0250: /* Task command name length */
0251: #define TASK_COMM_LEN 16
0252:
0253: #include <linux/spinlock.h>
0254:
0255: /*
0256: * This serializes "schedule()" and also protects
0257: * the run-queue from deletions/modifications (but
0258: * _adding_ to the beginning of the run-queue has
0259: * a separate lock).
0260: */
0261: extern rwlock_t tasklist_lock;
0262: extern spinlock_t mmlist_lock;
0263:
0264: struct task_struct;
0265:
0266: #ifdef CONFIG_PROVE_RCU
0267: extern int lockdep_tasklist_lock_is_held(void);
0268: #endif /* #ifdef CONFIG_PROVE_RCU */
0269:
0270: extern void sched_init(void);
0271: extern void sched_init_smp(void);
0272: extern asmlinkage void schedule_tail(struct task_struct *prev);
0273: extern void init_idle(struct task_struct *idle, int cpu);
0274: extern void init_idle_bootup_task(struct task_struct *idle);
0275:
0276: extern int runqueue_is_locked(int cpu);
0277:
0278: #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
0279: extern void select_nohz_load_balancer(int stop_tick);
0280: extern int get_nohz_timer_target(void);
0281: #else
0282: static inline void select_nohz_load_balancer(int stop_tick) { }
0283: #endif
0284:
0285: /*
0286: * Only dump TASK_* tasks. (0 for all tasks)
0287: */
0288: extern void show_state_filter(unsigned long state_filter);
0289:
0290: static inline void show_state(void)
0291: {
0292: show_state_filter(0);
0293: }
0294:
0295: extern void show_regs(struct pt_regs *);
0296:
0297: /*
0298: * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
0299: * task), SP is the stack pointer of the first frame that should be shown in the back
0300: * trace (or NULL if the entire call-chain of the task should be shown).
0301: */
0302: extern void show_stack(struct task_struct *task, unsigned long *sp);
0303:
0304: void io_schedule(void);
0305: long io_schedule_timeout(long timeout);
0306:
0307: extern void cpu_init (void);
0308: extern void trap_init(void);
0309: extern void update_process_times(int user);
0310: extern void scheduler_tick(void);
0311:
0312: extern void sched_show_task(struct task_struct *p);
0313:
0314: #ifdef CONFIG_LOCKUP_DETECTOR
0315: extern void touch_softlockup_watchdog(void);
0316: extern void touch_softlockup_watchdog_sync(void);
0317: extern void touch_all_softlockup_watchdogs(void);
0318: extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
0319: void __user *buffer,
0320: size_t *lenp, loff_t *ppos);
0321: extern unsigned int softlockup_panic;
0322: void lockup_detector_init(void);
0323: #else
0324: static inline void touch_softlockup_watchdog(void)
0325: {
0326: }
0327: static inline void touch_softlockup_watchdog_sync(void)
0328: {
0329: }
0330: static inline void touch_all_softlockup_watchdogs(void)
0331: {
0332: }
0333: static inline void lockup_detector_init(void)
0334: {
0335: }
0336: #endif
0337:
0338: #ifdef CONFIG_DETECT_HUNG_TASK
0339: extern unsigned int sysctl_hung_task_panic;
0340: extern unsigned long sysctl_hung_task_check_count;
0341: extern unsigned long sysctl_hung_task_timeout_secs;
0342: extern unsigned long sysctl_hung_task_warnings;
0343: extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
0344: void __user *buffer,
0345: size_t *lenp, loff_t *ppos);
0346: #else
0347: /* Avoid need for ifdefs elsewhere in the code */
0348: enum { sysctl_hung_task_timeout_secs = 0 };
0349: #endif
0350:
0351: /* Attach to any functions which should be ignored in wchan output. */
0352: #define __sched __attribute__((__section__(".sched.text")))
0353:
0354: /* Linker adds these: start and end of __sched functions */
0355: extern char __sched_text_start[], __sched_text_end[];
0356:
0357: /* Is this address in the __sched functions? */
0358: extern int in_sched_functions(unsigned long addr);
0359:
0360: #define MAX_SCHEDULE_TIMEOUT LONG_MAX
0361: extern signed long schedule_timeout(signed long timeout);
0362: extern signed long schedule_timeout_interruptible(signed long timeout);
0363: extern signed long schedule_timeout_killable(signed long timeout);
0364: extern signed long schedule_timeout_uninterruptible(signed long timeout);
0365: asmlinkage void schedule(void);
0366: extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
0367:
0368: struct nsproxy;
0369: struct user_namespace;
0370:
0371: /*
0372: * Default maximum number of active map areas, this limits the number of vmas
0373: * per mm struct. Users can overwrite this number by sysctl but there is a
0374: * problem.
0375: *
0376: * When a program's coredump is generated as ELF format, a section is created
0377: * per a vma. In ELF, the number of sections is represented in unsigned short.
0378: * This means the number of sections should be smaller than 65535 at coredump.
0379: * Because the kernel adds some informative sections to a image of program at
0380: * generating coredump, we need some margin. The number of extra sections is
0381: * 1-3 now and depends on arch. We use "5" as safe margin, here.
0382: */
0383: #define MAPCOUNT_ELF_CORE_MARGIN (5)
0384: #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
0385:
0386: extern int sysctl_max_map_count;
0387:
0388: #include <linux/aio.h>
0389:
0390: #ifdef CONFIG_MMU
0391: extern void arch_pick_mmap_layout(struct mm_struct *mm);
0392: extern unsigned long
0393: arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
0394: unsigned long, unsigned long);
0395:
0396: extern unsigned long
0397: arch_get_unmapped_exec_area(struct file *, unsigned long, unsigned long,
0398: unsigned long, unsigned long);
0399: extern unsigned long
0400: arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
0401: unsigned long len, unsigned long pgoff,
0402: unsigned long flags);
0403: extern void arch_unmap_area(struct mm_struct *, unsigned long);
0404: extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
0405: #else
0406: static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
0407: #endif
0408:
0409:
0410: extern void set_dumpable(struct mm_struct *mm, int value);
0411: extern int get_dumpable(struct mm_struct *mm);
0412:
0413: /* mm flags */
0414: /* dumpable bits */
0415: #define MMF_DUMPABLE 0 /* core dump is permitted */
0416: #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
0417:
0418: #define MMF_DUMPABLE_BITS 2
0419: #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
0420:
0421: /* coredump filter bits */
0422: #define MMF_DUMP_ANON_PRIVATE 2
0423: #define MMF_DUMP_ANON_SHARED 3
0424: #define MMF_DUMP_MAPPED_PRIVATE 4
0425: #define MMF_DUMP_MAPPED_SHARED 5
0426: #define MMF_DUMP_ELF_HEADERS 6
0427: #define MMF_DUMP_HUGETLB_PRIVATE 7
0428: #define MMF_DUMP_HUGETLB_SHARED 8
0429:
0430: #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
0431: #define MMF_DUMP_FILTER_BITS 7
0432: #define MMF_DUMP_FILTER_MASK \
0433: (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
0434: #define MMF_DUMP_FILTER_DEFAULT \
0435: ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
0436: (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
0437:
0438: #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
0439: # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
0440: #else
0441: # define MMF_DUMP_MASK_DEFAULT_ELF 0
0442: #endif
0443: /* leave room for more dump flags */
0444: #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
0445: #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
0446:
0447: #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
0448:
0449: struct sighand_struct {
0450: atomic_t count;
0451: struct k_sigaction action[_NSIG];
0452: spinlock_t siglock;
0453: wait_queue_head_t signalfd_wqh;
0454: };
0455:
0456: struct pacct_struct {
0457: int ac_flag;
0458: long ac_exitcode;
0459: unsigned long ac_mem;
0460: cputime_t ac_utime, ac_stime;
0461: unsigned long ac_minflt, ac_majflt;
0462: };
0463:
0464: struct cpu_itimer {
0465: cputime_t expires;
0466: cputime_t incr;
0467: u32 error;
0468: u32 incr_error;
0469: };
0470:
0471: /**
0472: * struct task_cputime - collected CPU time counts
0473: * @utime: time spent in user mode, in &cputime_t units
0474: * @stime: time spent in kernel mode, in &cputime_t units
0475: * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
0476: *
0477: * This structure groups together three kinds of CPU time that are
0478: * tracked for threads and thread groups. Most things considering
0479: * CPU time want to group these counts together and treat all three
0480: * of them in parallel.
0481: */
0482: struct task_cputime {
0483: cputime_t utime;
0484: cputime_t stime;
0485: unsigned long long sum_exec_runtime;
0486: };
0487: /* Alternate field names when used to cache expirations. */
0488: #define prof_exp stime
0489: #define virt_exp utime
0490: #define sched_exp sum_exec_runtime
0491:
0492: #define INIT_CPUTIME \
0493: (struct task_cputime) { \
0494: .utime = cputime_zero, \
0495: .stime = cputime_zero, \
0496: .sum_exec_runtime = 0, \
0497: }
0498:
0499: /*
0500: * Disable preemption until the scheduler is running.
0501: * Reset by start_kernel()->sched_init()->init_idle().
0502: *
0503: * We include PREEMPT_ACTIVE to avoid cond_resched() from working
0504: * before the scheduler is active -- see should_resched().
0505: */
0506: #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
0507:
0508: /**
0509: * struct thread_group_cputimer - thread group interval timer counts
0510: * @cputime: thread group interval timers.
0511: * @running: non-zero when there are timers running and
0512: * @cputime receives updates.
0513: * @lock: lock for fields in this struct.
0514: *
0515: * This structure contains the version of task_cputime, above, that is
0516: * used for thread group CPU timer calculations.
0517: */
0518: struct thread_group_cputimer {
0519: struct task_cputime cputime;
0520: int running;
0521: raw_spinlock_t lock;
0522: };
0523:
0524: #include <linux/rwsem.h>
0525: struct autogroup;
0526:
0527: /*
0528: * NOTE! "signal_struct" does not have its own
0529: * locking, because a shared signal_struct always
0530: * implies a shared sighand_struct, so locking
0531: * sighand_struct is always a proper superset of
0532: * the locking of signal_struct.
0533: */
0534: struct signal_struct {
0535: atomic_t sigcnt;
0536: atomic_t live;
0537: int nr_threads;
0538:
0539: wait_queue_head_t wait_chldexit; /* for wait4() */
0540:
0541: /* current thread group signal load-balancing target: */
0542: struct task_struct *curr_target;
0543:
0544: /* shared signal handling: */
0545: struct sigpending shared_pending;
0546:
0547: /* thread group exit support */
0548: int group_exit_code;
0549: /* overloaded:
0550: * - notify group_exit_task when ->count is equal to notify_count
0551: * - everyone except group_exit_task is stopped during signal delivery
0552: * of fatal signals, group_exit_task processes the signal.
0553: */
0554: int notify_count;
0555: struct task_struct *group_exit_task;
0556:
0557: /* thread group stop support, overloads group_exit_code too */
0558: int group_stop_count;
0559: unsigned int flags; /* see SIGNAL_* flags below */
0560:
0561: /* POSIX.1b Interval Timers */
0562: struct list_head posix_timers;
0563:
0564: /* ITIMER_REAL timer for the process */
0565: struct hrtimer real_timer;
0566: struct pid *leader_pid;
0567: ktime_t it_real_incr;
0568:
0569: /*
0570: * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
0571: * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
0572: * values are defined to 0 and 1 respectively
0573: */
0574: struct cpu_itimer it[2];
0575:
0576: /*
0577: * Thread group totals for process CPU timers.
0578: * See thread_group_cputimer(), et al, for details.
0579: */
0580: struct thread_group_cputimer cputimer;
0581:
0582: /* Earliest-expiration cache. */
0583: struct task_cputime cputime_expires;
0584:
0585: struct list_head cpu_timers[3];
0586:
0587: struct pid *tty_old_pgrp;
0588:
0589: /* boolean value for session group leader */
0590: int leader;
0591:
0592: struct tty_struct *tty; /* NULL if no tty */
0593:
0594: #ifdef CONFIG_SCHED_AUTOGROUP
0595: struct autogroup *autogroup;
0596: #endif
0597: /*
0598: * Cumulative resource counters for dead threads in the group,
0599: * and for reaped dead child processes forked by this group.
0600: * Live threads maintain their own counters and add to these
0601: * in __exit_signal, except for the group leader.
0602: */
0603: cputime_t utime, stime, cutime, cstime;
0604: cputime_t gtime;
0605: cputime_t cgtime;
0606: #ifndef CONFIG_VIRT_CPU_ACCOUNTING
0607: cputime_t prev_utime, prev_stime;
0608: #endif
0609: unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
0610: unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
0611: unsigned long inblock, oublock, cinblock, coublock;
0612: unsigned long maxrss, cmaxrss;
0613: struct task_io_accounting ioac;
0614:
0615: /*
0616: * Cumulative ns of schedule CPU time fo dead threads in the
0617: * group, not including a zombie group leader, (This only differs
0618: * from jiffies_to_ns(utime + stime) if sched_clock uses something
0619: * other than jiffies.)
0620: */
0621: unsigned long long sum_sched_runtime;
0622:
0623: /*
0624: * We don't bother to synchronize most readers of this at all,
0625: * because there is no reader checking a limit that actually needs
0626: * to get both rlim_cur and rlim_max atomically, and either one
0627: * alone is a single word that can safely be read normally.
0628: * getrlimit/setrlimit use task_lock(current->group_leader) to
0629: * protect this instead of the siglock, because they really
0630: * have no need to disable irqs.
0631: */
0632: struct rlimit rlim[RLIM_NLIMITS];
0633:
0634: #ifdef CONFIG_BSD_PROCESS_ACCT
0635: struct pacct_struct pacct; /* per-process accounting information */
0636: #endif
0637: #ifdef CONFIG_TASKSTATS
0638: struct taskstats *stats;
0639: #endif
0640: #ifdef CONFIG_AUDIT
0641: unsigned audit_tty;
0642: struct tty_audit_buf *tty_audit_buf;
0643: #endif
0644: #ifdef CONFIG_CGROUPS
0645: /*
0646: * The threadgroup_fork_lock prevents threads from forking with
0647: * CLONE_THREAD while held for writing. Use this for fork-sensitive
0648: * threadgroup-wide operations. It's taken for reading in fork.c in
0649: * copy_process().
0650: * Currently only needed write-side by cgroups.
0651: */
0652: struct rw_semaphore threadgroup_fork_lock;
0653: #endif
0654:
0655: int oom_adj; /* OOM kill score adjustment (bit shift) */
0656: int oom_score_adj; /* OOM kill score adjustment */
0657: int oom_score_adj_min; /* OOM kill score adjustment minimum value.
0658: * Only settable by CAP_SYS_RESOURCE. */
0659:
0660: struct mutex cred_guard_mutex; /* guard against foreign influences on
0661: * credential calculations
0662: * (notably. ptrace) */
0663: };
0664:
0665: /* Context switch must be unlocked if interrupts are to be enabled */
0666: #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
0667: # define __ARCH_WANT_UNLOCKED_CTXSW
0668: #endif
0669:
0670: /*
0671: * Bits in flags field of signal_struct.
0672: */
0673: #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
0674: #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
0675: #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
0676: /*
0677: * Pending notifications to parent.
0678: */
0679: #define SIGNAL_CLD_STOPPED 0x00000010
0680: #define SIGNAL_CLD_CONTINUED 0x00000020
0681: #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
0682:
0683: #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
0684:
0685: /* If true, all threads except ->group_exit_task have pending SIGKILL */
0686: static inline int signal_group_exit(const struct signal_struct *sig)
0687: {
0688: return (sig->flags & SIGNAL_GROUP_EXIT) ||
0689: (sig->group_exit_task != NULL);
0690: }
0691:
0692: /*
0693: * Some day this will be a full-fledged user tracking system..
0694: */
0695: struct user_struct {
0696: atomic_t __count; /* reference count */
0697: atomic_t processes; /* How many processes does this user have? */
0698: atomic_t files; /* How many open files does this user have? */
0699: atomic_t sigpending; /* How many pending signals does this user have? */
0700: #ifdef CONFIG_INOTIFY_USER
0701: atomic_t inotify_watches; /* How many inotify watches does this user have? */
0702: atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
0703: #endif
0704: #ifdef CONFIG_FANOTIFY
0705: atomic_t fanotify_listeners;
0706: #endif
0707: #ifdef CONFIG_EPOLL
0708: atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
0709: #endif
0710: #ifdef CONFIG_POSIX_MQUEUE
0711: /* protected by mq_lock */
0712: unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
0713: #endif
0714: unsigned long locked_shm; /* How many pages of mlocked shm ? */
0715:
0716: #ifdef CONFIG_KEYS
0717: struct key *uid_keyring; /* UID specific keyring */
0718: struct key *session_keyring; /* UID's default session keyring */
0719: #endif
0720:
0721: /* Hash table maintenance information */
0722: struct hlist_node uidhash_node;
0723: uid_t uid;
0724: struct user_namespace *user_ns;
0725:
0726: #ifdef CONFIG_PERF_EVENTS
0727: atomic_long_t locked_vm;
0728: #endif
0729: };
0730:
0731: extern int uids_sysfs_init(void);
0732:
0733: extern struct user_struct *find_user(uid_t);
0734:
0735: extern struct user_struct root_user;
0736: #define INIT_USER (&root_user)
0737:
0738:
0739: struct backing_dev_info;
0740: struct reclaim_state;
0741:
0742: #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
0743: struct sched_info {
0744: /* cumulative counters */
0745: unsigned long pcount; /* # of times run on this cpu */
0746: unsigned long long run_delay; /* time spent waiting on a runqueue */
0747:
0748: /* timestamps */
0749: unsigned long long last_arrival,/* when we last ran on a cpu */
0750: last_queued; /* when we were last queued to run */
0751: };
0752: #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
0753:
0754: #ifdef CONFIG_TASK_DELAY_ACCT
0755: struct task_delay_info {
0756: spinlock_t lock;
0757: unsigned int flags; /* Private per-task flags */
0758:
0759: /* For each stat XXX, add following, aligned appropriately
0760: *
0761: * struct timespec XXX_start, XXX_end;
0762: * u64 XXX_delay;
0763: * u32 XXX_count;
0764: *
0765: * Atomicity of updates to XXX_delay, XXX_count protected by
0766: * single lock above (split into XXX_lock if contention is an issue).
0767: */
0768:
0769: /*
0770: * XXX_count is incremented on every XXX operation, the delay
0771: * associated with the operation is added to XXX_delay.
0772: * XXX_delay contains the accumulated delay time in nanoseconds.
0773: */
0774: struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
0775: u64 blkio_delay; /* wait for sync block io completion */
0776: u64 swapin_delay; /* wait for swapin block io completion */
0777: u32 blkio_count; /* total count of the number of sync block */
0778: /* io operations performed */
0779: u32 swapin_count; /* total count of the number of swapin block */
0780: /* io operations performed */
0781:
0782: struct timespec freepages_start, freepages_end;
0783: u64 freepages_delay; /* wait for memory reclaim */
0784: u32 freepages_count; /* total count of memory reclaim */
0785: };
0786: #endif /* CONFIG_TASK_DELAY_ACCT */
0787:
0788: static inline int sched_info_on(void)
0789: {
0790: #ifdef CONFIG_SCHEDSTATS
0791: return 1;
0792: #elif defined(CONFIG_TASK_DELAY_ACCT)
0793: extern int delayacct_on;
0794: return delayacct_on;
0795: #else
0796: return 0;
0797: #endif
0798: }
0799:
0800: enum cpu_idle_type {
0801: CPU_IDLE,
0802: CPU_NOT_IDLE,
0803: CPU_NEWLY_IDLE,
0804: CPU_MAX_IDLE_TYPES
0805: };
0806:
0807: /*
0808: * Increase resolution of nice-level calculations for 64-bit architectures.
0809: * The extra resolution improves shares distribution and load balancing of
0810: * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
0811: * hierarchies, especially on larger systems. This is not a user-visible change
0812: * and does not change the user-interface for setting shares/weights.
0813: *
0814: * We increase resolution only if we have enough bits to allow this increased
0815: * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
0816: * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
0817: * increased costs.
0818: */
0819: #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load */
0820: # define SCHED_LOAD_RESOLUTION 10
0821: # define scale_load(w) ((w) << SCHED_LOAD_RESOLUTION)
0822: # define scale_load_down(w) ((w) >> SCHED_LOAD_RESOLUTION)
0823: #else
0824: # define SCHED_LOAD_RESOLUTION 0
0825: # define scale_load(w) (w)
0826: # define scale_load_down(w) (w)
0827: #endif
0828:
0829: #define SCHED_LOAD_SHIFT (10 + SCHED_LOAD_RESOLUTION)
0830: #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
0831:
0832: /*
0833: * Increase resolution of cpu_power calculations
0834: */
0835: #define SCHED_POWER_SHIFT 10
0836: #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
0837:
0838: /*
0839: * sched-domains (multiprocessor balancing) declarations:
0840: */
0841: #ifdef CONFIG_SMP
0842: #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
0843: #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
0844: #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
0845: #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
0846: #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
0847: #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
0848: #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
0849: #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
0850: #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
0851: #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
0852: #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
0853: #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
0854: #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
0855: #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
0856:
0857: enum powersavings_balance_level {
0858: POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
0859: POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
0860: * first for long running threads
0861: */
0862: POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
0863: * cpu package for power savings
0864: */
0865: MAX_POWERSAVINGS_BALANCE_LEVELS
0866: };
0867:
0868: extern int sched_mc_power_savings, sched_smt_power_savings;
0869:
0870: static inline int sd_balance_for_mc_power(void)
0871: {
0872: if (sched_smt_power_savings)
0873: return SD_POWERSAVINGS_BALANCE;
0874:
0875: if (!sched_mc_power_savings)
0876: return SD_PREFER_SIBLING;
0877:
0878: return 0;
0879: }
0880:
0881: static inline int sd_balance_for_package_power(void)
0882: {
0883: if (sched_mc_power_savings | sched_smt_power_savings)
0884: return SD_POWERSAVINGS_BALANCE;
0885:
0886: return SD_PREFER_SIBLING;
0887: }
0888:
0889: extern int __weak arch_sd_sibiling_asym_packing(void);
0890:
0891: /*
0892: * Optimise SD flags for power savings:
0893: * SD_BALANCE_NEWIDLE helps aggressive task consolidation and power savings.
0894: * Keep default SD flags if sched_{smt,mc}_power_saving=0
0895: */
0896:
0897: static inline int sd_power_saving_flags(void)
0898: {
0899: if (sched_mc_power_savings | sched_smt_power_savings)
0900: return SD_BALANCE_NEWIDLE;
0901:
0902: return 0;
0903: }
0904:
0905: struct sched_group_power {
0906: atomic_t ref;
0907: /*
0908: * CPU power of this group, SCHED_LOAD_SCALE being max power for a
0909: * single CPU.
0910: */
0911: unsigned int power, power_orig;
0912: };
0913:
0914: struct sched_group {
0915: struct sched_group *next; /* Must be a circular list */
0916: atomic_t ref;
0917:
0918: unsigned int group_weight;
0919: struct sched_group_power *sgp;
0920:
0921: /*
0922: * The CPUs this group covers.
0923: *
0924: * NOTE: this field is variable length. (Allocated dynamically
0925: * by attaching extra space to the end of the structure,
0926: * depending on how many CPUs the kernel has booted up with)
0927: */
0928: unsigned long cpumask[0];
0929: };
0930:
0931: static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
0932: {
0933: return to_cpumask(sg->cpumask);
0934: }
0935:
0936: struct sched_domain_attr {
0937: int relax_domain_level;
0938: };
0939:
0940: #define SD_ATTR_INIT (struct sched_domain_attr) { \
0941: .relax_domain_level = -1, \
0942: }
0943:
0944: extern int sched_domain_level_max;
0945:
0946: struct sched_domain {
0947: /* These fields must be setup */
0948: struct sched_domain *parent; /* top domain must be null terminated */
0949: struct sched_domain *child; /* bottom domain must be null terminated */
0950: struct sched_group *groups; /* the balancing groups of the domain */
0951: unsigned long min_interval; /* Minimum balance interval ms */
0952: unsigned long max_interval; /* Maximum balance interval ms */
0953: unsigned int busy_factor; /* less balancing by factor if busy */
0954: unsigned int imbalance_pct; /* No balance until over watermark */
0955: unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
0956: unsigned int busy_idx;
0957: unsigned int idle_idx;
0958: unsigned int newidle_idx;
0959: unsigned int wake_idx;
0960: unsigned int forkexec_idx;
0961: unsigned int smt_gain;
0962: int flags; /* See SD_* */
0963: int level;
0964:
0965: /* Runtime fields. */
0966: unsigned long last_balance; /* init to jiffies. units in jiffies */
0967: unsigned int balance_interval; /* initialise to 1. units in ms. */
0968: unsigned int nr_balance_failed; /* initialise to 0 */
0969:
0970: u64 last_update;
0971:
0972: #ifdef CONFIG_SCHEDSTATS
0973: /* load_balance() stats */
0974: unsigned int lb_count[CPU_MAX_IDLE_TYPES];
0975: unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
0976: unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
0977: unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
0978: unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
0979: unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
0980: unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
0981: unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
0982:
0983: /* Active load balancing */
0984: unsigned int alb_count;
0985: unsigned int alb_failed;
0986: unsigned int alb_pushed;
0987:
0988: /* SD_BALANCE_EXEC stats */
0989: unsigned int sbe_count;
0990: unsigned int sbe_balanced;
0991: unsigned int sbe_pushed;
0992:
0993: /* SD_BALANCE_FORK stats */
0994: unsigned int sbf_count;
0995: unsigned int sbf_balanced;
0996: unsigned int sbf_pushed;
0997:
0998: /* try_to_wake_up() stats */
0999: unsigned int ttwu_wake_remote;
1000: unsigned int ttwu_move_affine;
1001: unsigned int ttwu_move_balance;
1002: #endif
1003: #ifdef CONFIG_SCHED_DEBUG
1004: char *name;
1005: #endif
1006: union {
1007: void *private; /* used during construction */
1008: struct rcu_head rcu; /* used during destruction */
1009: };
1010:
1011: unsigned int span_weight;
1012: /*
1013: * Span of all CPUs in this domain.
1014: *
1015: * NOTE: this field is variable length. (Allocated dynamically
1016: * by attaching extra space to the end of the structure,
1017: * depending on how many CPUs the kernel has booted up with)
1018: */
1019: unsigned long span[0];
1020: };
1021:
1022: static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1023: {
1024: return to_cpumask(sd->span);
1025: }
1026:
1027: extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1028: struct sched_domain_attr *dattr_new);
1029:
1030: /* Allocate an array of sched domains, for partition_sched_domains(). */
1031: cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1032: void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1033:
1034: /* Test a flag in parent sched domain */
1035: static inline int test_sd_parent(struct sched_domain *sd, int flag)
1036: {
1037: if (sd->parent && (sd->parent->flags & flag))
1038: return 1;
1039:
1040: return 0;
1041: }
1042:
1043: unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1044: unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1045:
1046: #else /* CONFIG_SMP */
1047:
1048: struct sched_domain_attr;
1049:
1050: static inline void
1051: partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1052: struct sched_domain_attr *dattr_new)
1053: {
1054: }
1055: #endif /* !CONFIG_SMP */
1056:
1057:
1058: struct io_context; /* See blkdev.h */
1059:
1060:
1061: #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1062: extern void prefetch_stack(struct task_struct *t);
1063: #else
1064: static inline void prefetch_stack(struct task_struct *t) { }
1065: #endif
1066:
1067: struct audit_context; /* See audit.c */
1068: struct mempolicy;
1069: struct pipe_inode_info;
1070: struct uts_namespace;
1071:
1072: struct rq;
1073: struct sched_domain;
1074:
1075: /*
1076: * wake flags
1077: */
1078: #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1079: #define WF_FORK 0x02 /* child wakeup after fork */
1080: #define WF_MIGRATED 0x04 /* internal use, task got migrated */
1081:
1082: #define ENQUEUE_WAKEUP 1
1083: #define ENQUEUE_HEAD 2
1084: #ifdef CONFIG_SMP
1085: #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */
1086: #else
1087: #define ENQUEUE_WAKING 0
1088: #endif
1089:
1090: #define DEQUEUE_SLEEP 1
1091:
1092: struct sched_class {
1093: const struct sched_class *next;
1094:
1095: void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1096: void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1097: void (*yield_task) (struct rq *rq);
1098: bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1099:
1100: void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1101:
1102: struct task_struct * (*pick_next_task) (struct rq *rq);
1103: void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1104:
1105: #ifdef CONFIG_SMP
1106: int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1107:
1108: void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1109: void (*post_schedule) (struct rq *this_rq);
1110: void (*task_waking) (struct task_struct *task);
1111: void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1112:
1113: void (*set_cpus_allowed)(struct task_struct *p,
1114: const struct cpumask *newmask);
1115:
1116: void (*rq_online)(struct rq *rq);
1117: void (*rq_offline)(struct rq *rq);
1118: #endif
1119:
1120: void (*set_curr_task) (struct rq *rq);
1121: void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1122: void (*task_fork) (struct task_struct *p);
1123:
1124: void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1125: void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1126: void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1127: int oldprio);
1128:
1129: unsigned int (*get_rr_interval) (struct rq *rq,
1130: struct task_struct *task);
1131:
1132: #ifdef CONFIG_FAIR_GROUP_SCHED
1133: void (*task_move_group) (struct task_struct *p, int on_rq);
1134: #endif
1135: };
1136:
1137: struct load_weight {
1138: unsigned long weight, inv_weight;
1139: };
1140:
1141: #ifdef CONFIG_SCHEDSTATS
1142: struct sched_statistics {
1143: u64 wait_start;
1144: u64 wait_max;
1145: u64 wait_count;
1146: u64 wait_sum;
1147: u64 iowait_count;
1148: u64 iowait_sum;
1149:
1150: u64 sleep_start;
1151: u64 sleep_max;
1152: s64 sum_sleep_runtime;
1153:
1154: u64 block_start;
1155: u64 block_max;
1156: u64 exec_max;
1157: u64 slice_max;
1158:
1159: u64 nr_migrations_cold;
1160: u64 nr_failed_migrations_affine;
1161: u64 nr_failed_migrations_running;
1162: u64 nr_failed_migrations_hot;
1163: u64 nr_forced_migrations;
1164:
1165: u64 nr_wakeups;
1166: u64 nr_wakeups_sync;
1167: u64 nr_wakeups_migrate;
1168: u64 nr_wakeups_local;
1169: u64 nr_wakeups_remote;
1170: u64 nr_wakeups_affine;
1171: u64 nr_wakeups_affine_attempts;
1172: u64 nr_wakeups_passive;
1173: u64 nr_wakeups_idle;
1174: };
1175: #endif
1176:
1177: struct sched_entity {
1178: struct load_weight load; /* for load-balancing */
1179: struct rb_node run_node;
1180: struct list_head group_node;
1181: unsigned int on_rq;
1182:
1183: u64 exec_start;
1184: u64 sum_exec_runtime;
1185: u64 vruntime;
1186: u64 prev_sum_exec_runtime;
1187:
1188: u64 nr_migrations;
1189:
1190: #ifdef CONFIG_SCHEDSTATS
1191: struct sched_statistics statistics;
1192: #endif
1193:
1194: #ifdef CONFIG_FAIR_GROUP_SCHED
1195: struct sched_entity *parent;
1196: /* rq on which this entity is (to be) queued: */
1197: struct cfs_rq *cfs_rq;
1198: /* rq "owned" by this entity/group: */
1199: struct cfs_rq *my_q;
1200: #endif
1201: };
1202:
1203: struct sched_rt_entity {
1204: struct list_head run_list;
1205: unsigned long timeout;
1206: unsigned int time_slice;
1207: int nr_cpus_allowed;
1208:
1209: struct sched_rt_entity *back;
1210: #ifdef CONFIG_RT_GROUP_SCHED
1211: struct sched_rt_entity *parent;
1212: /* rq on which this entity is (to be) queued: */
1213: struct rt_rq *rt_rq;
1214: /* rq "owned" by this entity/group: */
1215: struct rt_rq *my_q;
1216: #endif
1217: };
1218:
1219: struct rcu_node;
1220:
1221: enum perf_event_task_context {
1222: perf_invalid_context = -1,
1223: perf_hw_context = 0,
1224: perf_sw_context,
1225: perf_nr_task_contexts,
1226: };
1227:
1228: struct task_struct {
1229: volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1230: void *stack;
1231: atomic_t usage;
1232: unsigned int flags; /* per process flags, defined below */
1233: unsigned int ptrace;
1234:
1235: #ifdef CONFIG_SMP
1236: struct llist_node wake_entry;
1237: int on_cpu;
1238: #endif
1239: int on_rq;
1240:
1241: int prio, static_prio, normal_prio;
1242: unsigned int rt_priority;
1243: const struct sched_class *sched_class;
1244: struct sched_entity se;
1245: struct sched_rt_entity rt;
1246: #ifdef CONFIG_CGROUP_SCHED
1247: struct task_group *sched_task_group;
1248: #endif
1249:
1250: #ifdef CONFIG_PREEMPT_NOTIFIERS
1251: /* list of struct preempt_notifier: */
1252: struct hlist_head preempt_notifiers;
1253: #endif
1254:
1255: /*
1256: * fpu_counter contains the number of consecutive context switches
1257: * that the FPU is used. If this is over a threshold, the lazy fpu
1258: * saving becomes unlazy to save the trap. This is an unsigned char
1259: * so that after 256 times the counter wraps and the behavior turns
1260: * lazy again; this to deal with bursty apps that only use FPU for
1261: * a short time
1262: */
1263: unsigned char fpu_counter;
1264: #ifdef CONFIG_BLK_DEV_IO_TRACE
1265: unsigned int btrace_seq;
1266: #endif
1267:
1268: unsigned int policy;
1269: cpumask_t cpus_allowed;
1270:
1271: #ifdef CONFIG_PREEMPT_RCU
1272: int rcu_read_lock_nesting;
1273: char rcu_read_unlock_special;
1274: struct list_head rcu_node_entry;
1275: #endif /* #ifdef CONFIG_PREEMPT_RCU */
1276: #ifdef CONFIG_TREE_PREEMPT_RCU
1277: struct rcu_node *rcu_blocked_node;
1278: #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1279: #ifdef CONFIG_RCU_BOOST
1280: struct rt_mutex *rcu_boost_mutex;
1281: #endif /* #ifdef CONFIG_RCU_BOOST */
1282:
1283: #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1284: struct sched_info sched_info;
1285: #endif
1286:
1287: struct list_head tasks;
1288: #ifdef CONFIG_SMP
1289: struct plist_node pushable_tasks;
1290: #endif
1291:
1292: struct mm_struct *mm, *active_mm;
1293: #ifdef CONFIG_COMPAT_BRK
1294: unsigned brk_randomized:1;
1295: #endif
1296: #if defined(SPLIT_RSS_COUNTING)
1297: struct task_rss_stat rss_stat;
1298: #endif
1299: /* task state */
1300: int exit_state;
1301: int exit_code, exit_signal;
1302: int pdeath_signal; /* The signal sent when the parent dies */
1303: unsigned int jobctl; /* JOBCTL_*, siglock protected */
1304: /* ??? */
1305: unsigned int personality;
1306: unsigned did_exec:1;
1307: unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1308: * execve */
1309: unsigned in_iowait:1;
1310:
1311: /* task may not gain privileges */
1312: unsigned no_new_privs:1;
1313:
1314: /* Revert to default priority/policy when forking */
1315: unsigned sched_reset_on_fork:1;
1316: unsigned sched_contributes_to_load:1;
1317:
1318: pid_t pid;
1319: pid_t tgid;
1320:
1321: #ifdef CONFIG_CC_STACKPROTECTOR
1322: /* Canary value for the -fstack-protector gcc feature */
1323: unsigned long stack_canary;
1324: #endif
1325:
1326: /*
1327: * pointers to (original) parent process, youngest child, younger sibling,
1328: * older sibling, respectively. (p->father can be replaced with
1329: * p->real_parent->pid)
1330: */
1331: struct task_struct *real_parent; /* real parent process */
1332: struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1333: /*
1334: * children/sibling forms the list of my natural children
1335: */
1336: struct list_head children; /* list of my children */
1337: struct list_head sibling; /* linkage in my parent's children list */
1338: struct task_struct *group_leader; /* threadgroup leader */
1339:
1340: /*
1341: * ptraced is the list of tasks this task is using ptrace on.
1342: * This includes both natural children and PTRACE_ATTACH targets.
1343: * p->ptrace_entry is p's link on the p->parent->ptraced list.
1344: */
1345: struct list_head ptraced;
1346: struct list_head ptrace_entry;
1347:
1348: /* PID/PID hash table linkage. */
1349: struct pid_link pids[PIDTYPE_MAX];
1350: struct list_head thread_group;
1351:
1352: struct completion *vfork_done; /* for vfork() */
1353: int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1354: int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1355:
1356: cputime_t utime, stime, utimescaled, stimescaled;
1357: cputime_t gtime;
1358: #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1359: cputime_t prev_utime, prev_stime;
1360: #endif
1361: unsigned long nvcsw, nivcsw; /* context switch counts */
1362: struct timespec start_time; /* monotonic time */
1363: struct timespec real_start_time; /* boot based time */
1364: /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1365: unsigned long min_flt, maj_flt;
1366:
1367: struct task_cputime cputime_expires;
1368: struct list_head cpu_timers[3];
1369:
1370: /* process credentials */
1371: const struct cred __rcu *real_cred; /* objective and real subjective task
1372: * credentials (COW) */
1373: const struct cred __rcu *cred; /* effective (overridable) subjective task
1374: * credentials (COW) */
1375: struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1376:
1377: char comm[TASK_COMM_LEN]; /* executable name excluding path
1378: - access with [gs]et_task_comm (which lock
1379: it with task_lock())
1380: - initialized normally by setup_new_exec */
1381: /* file system info */
1382: int link_count, total_link_count;
1383: #ifdef CONFIG_SYSVIPC
1384: /* ipc stuff */
1385: struct sysv_sem sysvsem;
1386: #endif
1387: #ifdef CONFIG_DETECT_HUNG_TASK
1388: /* hung task detection */
1389: unsigned long last_switch_count;
1390: #endif
1391: /* CPU-specific state of this task */
1392: struct thread_struct thread;
1393: /* filesystem information */
1394: struct fs_struct *fs;
1395: /* open file information */
1396: struct files_struct *files;
1397: /* namespaces */
1398: struct nsproxy *nsproxy;
1399: /* signal handlers */
1400: struct signal_struct *signal;
1401: struct sighand_struct *sighand;
1402:
1403: sigset_t blocked, real_blocked;
1404: sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1405: struct sigpending pending;
1406:
1407: unsigned long sas_ss_sp;
1408: size_t sas_ss_size;
1409: int (*notifier)(void *priv);
1410: void *notifier_data;
1411: sigset_t *notifier_mask;
1412: struct audit_context *audit_context;
1413: #ifdef CONFIG_AUDITSYSCALL
1414: uid_t loginuid;
1415: unsigned int sessionid;
1416: #endif
1417: struct seccomp seccomp;
1418:
1419: /* Thread group tracking */
1420: u32 parent_exec_id;
1421: u32 self_exec_id;
1422: /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1423: * mempolicy */
1424: spinlock_t alloc_lock;
1425:
1426: #ifdef CONFIG_GENERIC_HARDIRQS
1427: /* IRQ handler threads */
1428: struct irqaction *irqaction;
1429: #endif
1430:
1431: /* Protection of the PI data structures: */
1432: raw_spinlock_t pi_lock;
1433:
1434: #ifdef CONFIG_RT_MUTEXES
1435: /* PI waiters blocked on a rt_mutex held by this task */
1436: struct plist_head pi_waiters;
1437: /* Deadlock detection and priority inheritance handling */
1438: struct rt_mutex_waiter *pi_blocked_on;
1439: #endif
1440:
1441: #ifdef CONFIG_DEBUG_MUTEXES
1442: /* mutex deadlock detection */
1443: struct mutex_waiter *blocked_on;
1444: #endif
1445: #ifdef CONFIG_TRACE_IRQFLAGS
1446: unsigned int irq_events;
1447: unsigned long hardirq_enable_ip;
1448: unsigned long hardirq_disable_ip;
1449: unsigned int hardirq_enable_event;
1450: unsigned int hardirq_disable_event;
1451: int hardirqs_enabled;
1452: int hardirq_context;
1453: unsigned long softirq_disable_ip;
1454: unsigned long softirq_enable_ip;
1455: unsigned int softirq_disable_event;
1456: unsigned int softirq_enable_event;
1457: int softirqs_enabled;
1458: int softirq_context;
1459: #endif
1460: #ifdef CONFIG_LOCKDEP
1461: # define MAX_LOCK_DEPTH 48UL
1462: u64 curr_chain_key;
1463: int lockdep_depth;
1464: unsigned int lockdep_recursion;
1465: struct held_lock held_locks[MAX_LOCK_DEPTH];
1466: gfp_t lockdep_reclaim_gfp;
1467: #endif
1468:
1469: /* journalling filesystem info */
1470: void *journal_info;
1471:
1472: /* stacked block device info */
1473: struct bio_list *bio_list;
1474:
1475: #ifdef CONFIG_BLOCK
1476: /* stack plugging */
1477: struct blk_plug *plug;
1478: #endif
1479:
1480: /* VM state */
1481: struct reclaim_state *reclaim_state;
1482:
1483: struct backing_dev_info *backing_dev_info;
1484:
1485: struct io_context *io_context;
1486:
1487: unsigned long ptrace_message;
1488: siginfo_t *last_siginfo; /* For ptrace use. */
1489: struct task_io_accounting ioac;
1490: #if defined(CONFIG_TASK_XACCT)
1491: u64 acct_rss_mem1; /* accumulated rss usage */
1492: u64 acct_vm_mem1; /* accumulated virtual memory usage */
1493: cputime_t acct_timexpd; /* stime + utime since last update */
1494: #endif
1495: #ifdef CONFIG_CPUSETS
1496: nodemask_t mems_allowed; /* Protected by alloc_lock */
1497: seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1498: int cpuset_mem_spread_rotor;
1499: int cpuset_slab_spread_rotor;
1500: #endif
1501: #ifdef CONFIG_CGROUPS
1502: /* Control Group info protected by css_set_lock */
1503: struct css_set __rcu *cgroups;
1504: /* cg_list protected by css_set_lock and tsk->alloc_lock */
1505: struct list_head cg_list;
1506: #endif
1507: #ifdef CONFIG_FUTEX
1508: struct robust_list_head __user *robust_list;
1509: #ifdef CONFIG_COMPAT
1510: struct compat_robust_list_head __user *compat_robust_list;
1511: #endif
1512: struct list_head pi_state_list;
1513: struct futex_pi_state *pi_state_cache;
1514: #endif
1515: #ifdef CONFIG_PERF_EVENTS
1516: struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1517: struct mutex perf_event_mutex;
1518: struct list_head perf_event_list;
1519: #endif
1520: #ifdef CONFIG_NUMA
1521: struct mempolicy *mempolicy; /* Protected by alloc_lock */
1522: short il_next;
1523: short pref_node_fork;
1524: #endif
1525: struct rcu_head rcu;
1526:
1527: /*
1528: * cache last used pipe for splice
1529: */
1530: struct pipe_inode_info *splice_pipe;
1531: #ifdef CONFIG_TASK_DELAY_ACCT
1532: struct task_delay_info *delays;
1533: #endif
1534: #ifdef CONFIG_FAULT_INJECTION
1535: int make_it_fail;
1536: #endif
1537: /*
1538: * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1539: * balance_dirty_pages() for some dirty throttling pause
1540: */
1541: int nr_dirtied;
1542: int nr_dirtied_pause;
1543:
1544: #ifdef CONFIG_LATENCYTOP
1545: int latency_record_count;
1546: struct latency_record latency_record[LT_SAVECOUNT];
1547: #endif
1548: /*
1549: * time slack values; these are used to round up poll() and
1550: * select() etc timeout values. These are in nanoseconds.
1551: */
1552: unsigned long timer_slack_ns;
1553: unsigned long default_timer_slack_ns;
1554:
1555: struct list_head *scm_work_list;
1556: #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1557: /* Index of current stored address in ret_stack */
1558: int curr_ret_stack;
1559: /* Stack of return addresses for return function tracing */
1560: struct ftrace_ret_stack *ret_stack;
1561: /* time stamp for last schedule */
1562: unsigned long long ftrace_timestamp;
1563: /*
1564: * Number of functions that haven't been traced
1565: * because of depth overrun.
1566: */
1567: atomic_t trace_overrun;
1568: /* Pause for the tracing */
1569: atomic_t tracing_graph_pause;
1570: #endif
1571: #ifdef CONFIG_TRACING
1572: /* state flags for use by tracers */
1573: unsigned long trace;
1574: /* bitmask and counter of trace recursion */
1575: unsigned long trace_recursion;
1576: #endif /* CONFIG_TRACING */
1577: #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1578: struct memcg_batch_info {
1579: int do_batch; /* incremented when batch uncharge started */
1580: struct mem_cgroup *memcg; /* target memcg of uncharge */
1581: unsigned long nr_pages; /* uncharged usage */
1582: unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1583: } memcg_batch;
1584: #endif
1585: #ifdef CONFIG_HAVE_HW_BREAKPOINT
1586: atomic_t ptrace_bp_refcnt;
1587: #endif
1588: };
1589:
1590: /* Future-safe accessor for struct task_struct's cpus_allowed. */
1591: #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1592:
1593: /*
1594: * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1595: * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1596: * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1597: * values are inverted: lower p->prio value means higher priority.
1598: *
1599: * The MAX_USER_RT_PRIO value allows the actual maximum
1600: * RT priority to be separate from the value exported to
1601: * user-space. This allows kernel threads to set their
1602: * priority to a value higher than any user task. Note:
1603: * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1604: */
1605:
1606: #define MAX_USER_RT_PRIO 100
1607: #define MAX_RT_PRIO MAX_USER_RT_PRIO
1608:
1609: #define MAX_PRIO (MAX_RT_PRIO + 40)
1610: #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1611:
1612: static inline int rt_prio(int prio)
1613: {
1614: if (unlikely(prio < MAX_RT_PRIO))
1615: return 1;
1616: return 0;
1617: }
1618:
1619: static inline int rt_task(struct task_struct *p)
1620: {
1621: return rt_prio(p->prio);
1622: }
1623:
1624: static inline struct pid *task_pid(struct task_struct *task)
1625: {
1626: return task->pids[PIDTYPE_PID].pid;
1627: }
1628:
1629: static inline struct pid *task_tgid(struct task_struct *task)
1630: {
1631: return task->group_leader->pids[PIDTYPE_PID].pid;
1632: }
1633:
1634: /*
1635: * Without tasklist or rcu lock it is not safe to dereference
1636: * the result of task_pgrp/task_session even if task == current,
1637: * we can race with another thread doing sys_setsid/sys_setpgid.
1638: */
1639: static inline struct pid *task_pgrp(struct task_struct *task)
1640: {
1641: return task->group_leader->pids[PIDTYPE_PGID].pid;
1642: }
1643:
1644: static inline struct pid *task_session(struct task_struct *task)
1645: {
1646: return task->group_leader->pids[PIDTYPE_SID].pid;
1647: }
1648:
1649: struct pid_namespace;
1650:
1651: /*
1652: * the helpers to get the task's different pids as they are seen
1653: * from various namespaces
1654: *
1655: * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1656: * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1657: * current.
1658: * task_xid_nr_ns() : id seen from the ns specified;
1659: *
1660: * set_task_vxid() : assigns a virtual id to a task;
1661: *
1662: * see also pid_nr() etc in include/linux/pid.h
1663: */
1664: pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1665: struct pid_namespace *ns);
1666:
1667: static inline pid_t task_pid_nr(struct task_struct *tsk)
1668: {
1669: return tsk->pid;
1670: }
1671:
1672: static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1673: struct pid_namespace *ns)
1674: {
1675: return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1676: }
1677:
1678: static inline pid_t task_pid_vnr(struct task_struct *tsk)
1679: {
1680: return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1681: }
1682:
1683:
1684: static inline pid_t task_tgid_nr(struct task_struct *tsk)
1685: {
1686: return tsk->tgid;
1687: }
1688:
1689: pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1690:
1691: static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1692: {
1693: return pid_vnr(task_tgid(tsk));
1694: }
1695:
1696:
1697: static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1698: struct pid_namespace *ns)
1699: {
1700: return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1701: }
1702:
1703: static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1704: {
1705: return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1706: }
1707:
1708:
1709: static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1710: struct pid_namespace *ns)
1711: {
1712: return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1713: }
1714:
1715: static inline pid_t task_session_vnr(struct task_struct *tsk)
1716: {
1717: return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1718: }
1719:
1720: /* obsolete, do not use */
1721: static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1722: {
1723: return task_pgrp_nr_ns(tsk, &init_pid_ns);
1724: }
1725:
1726: /**
1727: * pid_alive - check that a task structure is not stale
1728: * @p: Task structure to be checked.
1729: *
1730: * Test if a process is not yet dead (at most zombie state)
1731: * If pid_alive fails, then pointers within the task structure
1732: * can be stale and must not be dereferenced.
1733: */
1734: static inline int pid_alive(struct task_struct *p)
1735: {
1736: return p->pids[PIDTYPE_PID].pid != NULL;
1737: }
1738:
1739: /**
1740: * is_global_init - check if a task structure is init
1741: * @tsk: Task structure to be checked.
1742: *
1743: * Check if a task structure is the first user space task the kernel created.
1744: */
1745: static inline int is_global_init(struct task_struct *tsk)
1746: {
1747: return tsk->pid == 1;
1748: }
1749:
1750: /*
1751: * is_container_init:
1752: * check whether in the task is init in its own pid namespace.
1753: */
1754: extern int is_container_init(struct task_struct *tsk);
1755:
1756: extern struct pid *cad_pid;
1757:
1758: extern void free_task(struct task_struct *tsk);
1759: #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1760:
1761: extern void __put_task_struct(struct task_struct *t);
1762:
1763: static inline void put_task_struct(struct task_struct *t)
1764: {
1765: if (atomic_dec_and_test(&t->usage))
1766: __put_task_struct(t);
1767: }
1768:
1769: extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1770: extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1771:
1772: /*
1773: * Per process flags
1774: */
1775: #define PF_STARTING 0x00000002 /* being created */
1776: #define PF_EXITING 0x00000004 /* getting shut down */
1777: #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1778: #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1779: #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1780: #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1781: #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1782: #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1783: #define PF_DUMPCORE 0x00000200 /* dumped core */
1784: #define PF_SIGNALED 0x00000400 /* killed by a signal */
1785: #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1786: #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1787: #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1788: #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1789: #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1790: #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1791: #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1792: #define PF_KSWAPD 0x00040000 /* I am kswapd */
1793: #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1794: #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1795: #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1796: #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1797: #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1798: #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1799: #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1800: #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1801: #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1802: #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1803: #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1804: #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1805:
1806: /*
1807: * Only the _current_ task can read/write to tsk->flags, but other
1808: * tasks can access tsk->flags in readonly mode for example
1809: * with tsk_used_math (like during threaded core dumping).
1810: * There is however an exception to this rule during ptrace
1811: * or during fork: the ptracer task is allowed to write to the
1812: * child->flags of its traced child (same goes for fork, the parent
1813: * can write to the child->flags), because we're guaranteed the
1814: * child is not running and in turn not changing child->flags
1815: * at the same time the parent does it.
1816: */
1817: #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1818: #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1819: #define clear_used_math() clear_stopped_child_used_math(current)
1820: #define set_used_math() set_stopped_child_used_math(current)
1821: #define conditional_stopped_child_used_math(condition, child) \
1822: do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1823: #define conditional_used_math(condition) \
1824: conditional_stopped_child_used_math(condition, current)
1825: #define copy_to_stopped_child_used_math(child) \
1826: do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1827: /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1828: #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1829: #define used_math() tsk_used_math(current)
1830:
1831: /*
1832: * task->jobctl flags
1833: */
1834: #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1835:
1836: #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1837: #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1838: #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1839: #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1840: #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1841: #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1842: #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1843:
1844: #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1845: #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1846: #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1847: #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1848: #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1849: #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1850: #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1851:
1852: #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1853: #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1854:
1855: extern bool task_set_jobctl_pending(struct task_struct *task,
1856: unsigned int mask);
1857: extern void task_clear_jobctl_trapping(struct task_struct *task);
1858: extern void task_clear_jobctl_pending(struct task_struct *task,
1859: unsigned int mask);
1860:
1861: #ifdef CONFIG_PREEMPT_RCU
1862:
1863: #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1864: #define RCU_READ_UNLOCK_BOOSTED (1 << 1) /* boosted while in RCU read-side. */
1865: #define RCU_READ_UNLOCK_NEED_QS (1 << 2) /* RCU core needs CPU response. */
1866:
1867: static inline void rcu_copy_process(struct task_struct *p)
1868: {
1869: p->rcu_read_lock_nesting = 0;
1870: p->rcu_read_unlock_special = 0;
1871: #ifdef CONFIG_TREE_PREEMPT_RCU
1872: p->rcu_blocked_node = NULL;
1873: #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1874: #ifdef CONFIG_RCU_BOOST
1875: p->rcu_boost_mutex = NULL;
1876: #endif /* #ifdef CONFIG_RCU_BOOST */
1877: INIT_LIST_HEAD(&p->rcu_node_entry);
1878: }
1879:
1880: #else
1881:
1882: static inline void rcu_copy_process(struct task_struct *p)
1883: {
1884: }
1885:
1886: #endif
1887:
1888: #ifdef CONFIG_SMP
1889: extern void do_set_cpus_allowed(struct task_struct *p,
1890: const struct cpumask *new_mask);
1891:
1892: extern int set_cpus_allowed_ptr(struct task_struct *p,
1893: const struct cpumask *new_mask);
1894: #else
1895: static inline void do_set_cpus_allowed(struct task_struct *p,
1896: const struct cpumask *new_mask)
1897: {
1898: }
1899: static inline int set_cpus_allowed_ptr(struct task_struct *p,
1900: const struct cpumask *new_mask)
1901: {
1902: if (!cpumask_test_cpu(0, new_mask))
1903: return -EINVAL;
1904: return 0;
1905: }
1906: #endif
1907:
1908: #ifdef CONFIG_NO_HZ
1909: void calc_load_enter_idle(void);
1910: void calc_load_exit_idle(void);
1911: #else
1912: static inline void calc_load_enter_idle(void) { }
1913: static inline void calc_load_exit_idle(void) { }
1914: #endif /* CONFIG_NO_HZ */
1915:
1916: #ifndef CONFIG_CPUMASK_OFFSTACK
1917: static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1918: {
1919: return set_cpus_allowed_ptr(p, &new_mask);
1920: }
1921: #endif
1922:
1923: /*
1924: * Do not use outside of architecture code which knows its limitations.
1925: *
1926: * sched_clock() has no promise of monotonicity or bounded drift between
1927: * CPUs, use (which you should not) requires disabling IRQs.
1928: *
1929: * Please use one of the three interfaces below.
1930: */
1931: extern unsigned long long notrace sched_clock(void);
1932: /*
1933: * See the comment in kernel/sched_clock.c
1934: */
1935: extern u64 cpu_clock(int cpu);
1936: extern u64 local_clock(void);
1937: extern u64 sched_clock_cpu(int cpu);
1938:
1939:
1940: extern void sched_clock_init(void);
1941:
1942: #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1943: static inline void sched_clock_tick(void)
1944: {
1945: }
1946:
1947: static inline void sched_clock_idle_sleep_event(void)
1948: {
1949: }
1950:
1951: static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1952: {
1953: }
1954: #else
1955: /*
1956: * Architectures can set this to 1 if they have specified
1957: * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1958: * but then during bootup it turns out that sched_clock()
1959: * is reliable after all:
1960: */
1961: extern int sched_clock_stable;
1962:
1963: extern void sched_clock_tick(void);
1964: extern void sched_clock_idle_sleep_event(void);
1965: extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1966: #endif
1967:
1968: #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1969: /*
1970: * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1971: * The reason for this explicit opt-in is not to have perf penalty with
1972: * slow sched_clocks.
1973: */
1974: extern void enable_sched_clock_irqtime(void);
1975: extern void disable_sched_clock_irqtime(void);
1976: #else
1977: static inline void enable_sched_clock_irqtime(void) {}
1978: static inline void disable_sched_clock_irqtime(void) {}
1979: #endif
1980:
1981: extern unsigned long long
1982: task_sched_runtime(struct task_struct *task);
1983:
1984: /* sched_exec is called by processes performing an exec */
1985: #ifdef CONFIG_SMP
1986: extern void sched_exec(void);
1987: #else
1988: #define sched_exec() {}
1989: #endif
1990:
1991: extern void sched_clock_idle_sleep_event(void);
1992: extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1993:
1994: #ifdef CONFIG_HOTPLUG_CPU
1995: extern void idle_task_exit(void);
1996: #else
1997: static inline void idle_task_exit(void) {}
1998: #endif
1999:
2000: #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
2001: extern void wake_up_idle_cpu(int cpu);
2002: #else
2003: static inline void wake_up_idle_cpu(int cpu) { }
2004: #endif
2005:
2006: extern unsigned int sysctl_sched_latency;
2007: extern unsigned int sysctl_sched_min_granularity;
2008: extern unsigned int sysctl_sched_wakeup_granularity;
2009: extern unsigned int sysctl_sched_child_runs_first;
2010:
2011: enum sched_tunable_scaling {
2012: SCHED_TUNABLESCALING_NONE,
2013: SCHED_TUNABLESCALING_LOG,
2014: SCHED_TUNABLESCALING_LINEAR,
2015: SCHED_TUNABLESCALING_END,
2016: };
2017: extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
2018:
2019: #ifdef CONFIG_SCHED_DEBUG
2020: extern unsigned int sysctl_sched_migration_cost;
2021: extern unsigned int sysctl_sched_nr_migrate;
2022: extern unsigned int sysctl_sched_time_avg;
2023: extern unsigned int sysctl_timer_migration;
2024: extern unsigned int sysctl_sched_shares_window;
2025:
2026: int sched_proc_update_handler(struct ctl_table *table, int write,
2027: void __user *buffer, size_t *length,
2028: loff_t *ppos);
2029: #endif
2030: #ifdef CONFIG_SCHED_DEBUG
2031: static inline unsigned int get_sysctl_timer_migration(void)
2032: {
2033: return sysctl_timer_migration;
2034: }
2035: #else
2036: static inline unsigned int get_sysctl_timer_migration(void)
2037: {
2038: return 1;
2039: }
2040: #endif
2041: extern unsigned int sysctl_sched_rt_period;
2042: extern int sysctl_sched_rt_runtime;
2043:
2044: int sched_rt_handler(struct ctl_table *table, int write,
2045: void __user *buffer, size_t *lenp,
2046: loff_t *ppos);
2047:
2048: #ifdef CONFIG_SCHED_AUTOGROUP
2049: extern unsigned int sysctl_sched_autogroup_enabled;
2050:
2051: extern void sched_autogroup_create_attach(struct task_struct *p);
2052: extern void sched_autogroup_detach(struct task_struct *p);
2053: extern void sched_autogroup_fork(struct signal_struct *sig);
2054: extern void sched_autogroup_exit(struct signal_struct *sig);
2055: #ifdef CONFIG_PROC_FS
2056: extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2057: extern int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice);
2058: #endif
2059: #else
2060: static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2061: static inline void sched_autogroup_detach(struct task_struct *p) { }
2062: static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2063: static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2064: #endif
2065:
2066: #ifdef CONFIG_CFS_BANDWIDTH
2067: extern unsigned int sysctl_sched_cfs_bandwidth_slice;
2068: #endif
2069:
2070: #ifdef CONFIG_RT_MUTEXES
2071: extern int rt_mutex_getprio(struct task_struct *p);
2072: extern void rt_mutex_setprio(struct task_struct *p, int prio);
2073: extern void rt_mutex_adjust_pi(struct task_struct *p);
2074: #else
2075: static inline int rt_mutex_getprio(struct task_struct *p)
2076: {
2077: return p->normal_prio;
2078: }
2079: # define rt_mutex_adjust_pi(p) do { } while (0)
2080: #endif
2081:
2082: extern bool yield_to(struct task_struct *p, bool preempt);
2083: extern void set_user_nice(struct task_struct *p, long nice);
2084: extern int task_prio(const struct task_struct *p);
2085: extern int task_nice(const struct task_struct *p);
2086: extern int can_nice(const struct task_struct *p, const int nice);
2087: extern int task_curr(const struct task_struct *p);
2088: extern int idle_cpu(int cpu);
2089: extern int sched_setscheduler(struct task_struct *, int,
2090: const struct sched_param *);
2091: extern int sched_setscheduler_nocheck(struct task_struct *, int,
2092: const struct sched_param *);
2093: extern struct task_struct *idle_task(int cpu);
2094: extern struct task_struct *curr_task(int cpu);
2095: extern void set_curr_task(int cpu, struct task_struct *p);
2096:
2097: void yield(void);
2098:
2099: /*
2100: * The default (Linux) execution domain.
2101: */
2102: extern struct exec_domain default_exec_domain;
2103:
2104: union thread_union {
2105: struct thread_info thread_info;
2106: unsigned long stack[THREAD_SIZE/sizeof(long)];
2107: };
2108:
2109: #ifndef __HAVE_ARCH_KSTACK_END
2110: static inline int kstack_end(void *addr)
2111: {
2112: /* Reliable end of stack detection:
2113: * Some APM bios versions misalign the stack
2114: */
2115: return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2116: }
2117: #endif
2118:
2119: extern union thread_union init_thread_union;
2120: extern struct task_struct init_task;
2121:
2122: extern struct mm_struct init_mm;
2123:
2124: extern struct pid_namespace init_pid_ns;
2125:
2126: /*
2127: * find a task by one of its numerical ids
2128: *
2129: * find_task_by_pid_ns():
2130: * finds a task by its pid in the specified namespace
2131: * find_task_by_vpid():
2132: * finds a task by its virtual pid
2133: *
2134: * see also find_vpid() etc in include/linux/pid.h
2135: */
2136:
2137: extern struct task_struct *find_task_by_vpid(pid_t nr);
2138: extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2139: struct pid_namespace *ns);
2140:
2141: extern void __set_special_pids(struct pid *pid);
2142:
2143: /* per-UID process charging. */
2144: extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2145: static inline struct user_struct *get_uid(struct user_struct *u)
2146: {
2147: atomic_inc(&u->__count);
2148: return u;
2149: }
2150: extern void free_uid(struct user_struct *);
2151: extern void release_uids(struct user_namespace *ns);
2152:
2153: #include <asm/current.h>
2154:
2155: extern void xtime_update(unsigned long ticks);
2156:
2157: extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2158: extern int wake_up_process(struct task_struct *tsk);
2159: extern void wake_up_new_task(struct task_struct *tsk);
2160: #ifdef CONFIG_SMP
2161: extern void kick_process(struct task_struct *tsk);
2162: #else
2163: static inline void kick_process(struct task_struct *tsk) { }
2164: #endif
2165: extern void sched_fork(struct task_struct *p);
2166: extern void sched_dead(struct task_struct *p);
2167:
2168: extern void proc_caches_init(void);
2169: extern void flush_signals(struct task_struct *);
2170: extern void __flush_signals(struct task_struct *);
2171: extern void ignore_signals(struct task_struct *);
2172: extern void flush_signal_handlers(struct task_struct *, int force_default);
2173: extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2174:
2175: static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2176: {
2177: unsigned long flags;
2178: int ret;
2179:
2180: spin_lock_irqsave(&tsk->sighand->siglock, flags);
2181: ret = dequeue_signal(tsk, mask, info);
2182: spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2183:
2184: return ret;
2185: }
2186:
2187: extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2188: sigset_t *mask);
2189: extern void unblock_all_signals(void);
2190: extern void release_task(struct task_struct * p);
2191: extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2192: extern int force_sigsegv(int, struct task_struct *);
2193: extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2194: extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2195: extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2196: extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2197: const struct cred *, u32);
2198: extern int kill_pgrp(struct pid *pid, int sig, int priv);
2199: extern int kill_pid(struct pid *pid, int sig, int priv);
2200: extern int kill_proc_info(int, struct siginfo *, pid_t);
2201: extern __must_check bool do_notify_parent(struct task_struct *, int);
2202: extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2203: extern void force_sig(int, struct task_struct *);
2204: extern int send_sig(int, struct task_struct *, int);
2205: extern int zap_other_threads(struct task_struct *p);
2206: extern struct sigqueue *sigqueue_alloc(void);
2207: extern void sigqueue_free(struct sigqueue *);
2208: extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2209: extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2210: extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2211:
2212: static inline int kill_cad_pid(int sig, int priv)
2213: {
2214: return kill_pid(cad_pid, sig, priv);
2215: }
2216:
2217: /* These can be the second arg to send_sig_info/send_group_sig_info. */
2218: #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2219: #define SEND_SIG_PRIV ((struct siginfo *) 1)
2220: #define SEND_SIG_FORCED ((struct siginfo *) 2)
2221:
2222: /*
2223: * True if we are on the alternate signal stack.
2224: */
2225: static inline int on_sig_stack(unsigned long sp)
2226: {
2227: #ifdef CONFIG_STACK_GROWSUP
2228: return sp >= current->sas_ss_sp &&
2229: sp - current->sas_ss_sp < current->sas_ss_size;
2230: #else
2231: return sp > current->sas_ss_sp &&
2232: sp - current->sas_ss_sp <= current->sas_ss_size;
2233: #endif
2234: }
2235:
2236: static inline int sas_ss_flags(unsigned long sp)
2237: {
2238: return (current->sas_ss_size == 0 ? SS_DISABLE
2239: : on_sig_stack(sp) ? SS_ONSTACK : 0);
2240: }
2241:
2242: /*
2243: * Routines for handling mm_structs
2244: */
2245: extern struct mm_struct * mm_alloc(void);
2246:
2247: /* mmdrop drops the mm and the page tables */
2248: extern void __mmdrop(struct mm_struct *);
2249: static inline void mmdrop(struct mm_struct * mm)
2250: {
2251: if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2252: __mmdrop(mm);
2253: }
2254:
2255: /* mmput gets rid of the mappings and all user-space */
2256: extern void mmput(struct mm_struct *);
2257: /* Grab a reference to a task's mm, if it is not already going away */
2258: extern struct mm_struct *get_task_mm(struct task_struct *task);
2259: /* Remove the current tasks stale references to the old mm_struct */
2260: extern void mm_release(struct task_struct *, struct mm_struct *);
2261: /* Allocate a new mm structure and copy contents from tsk->mm */
2262: extern struct mm_struct *dup_mm(struct task_struct *tsk);
2263:
2264: extern int copy_thread(unsigned long, unsigned long, unsigned long,
2265: struct task_struct *, struct pt_regs *);
2266: extern void flush_thread(void);
2267: extern void exit_thread(void);
2268:
2269: extern void exit_files(struct task_struct *);
2270: extern void __cleanup_sighand(struct sighand_struct *);
2271:
2272: extern void exit_itimers(struct signal_struct *);
2273: extern void flush_itimer_signals(void);
2274:
2275: extern NORET_TYPE void do_group_exit(int);
2276:
2277: extern void daemonize(const char *, ...);
2278: extern int allow_signal(int);
2279: extern int disallow_signal(int);
2280:
2281: extern int do_execve(const char *,
2282: const char __user * const __user *,
2283: const char __user * const __user *, struct pt_regs *);
2284: extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2285: struct task_struct *fork_idle(int);
2286:
2287: extern void set_task_comm(struct task_struct *tsk, char *from);
2288: extern char *get_task_comm(char *to, struct task_struct *tsk);
2289:
2290: #ifdef CONFIG_SMP
2291: void scheduler_ipi(void);
2292: extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2293: #else
2294: static inline void scheduler_ipi(void) { }
2295: static inline unsigned long wait_task_inactive(struct task_struct *p,
2296: long match_state)
2297: {
2298: return 1;
2299: }
2300: #endif
2301:
2302: #define next_task(p) \
2303: list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2304:
2305: #define for_each_process(p) \
2306: for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2307:
2308: extern bool current_is_single_threaded(void);
2309:
2310: /*
2311: * Careful: do_each_thread/while_each_thread is a double loop so
2312: * 'break' will not work as expected - use goto instead.
2313: */
2314: #define do_each_thread(g, t) \
2315: for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2316:
2317: #define while_each_thread(g, t) \
2318: while ((t = next_thread(t)) != g)
2319:
2320: static inline int get_nr_threads(struct task_struct *tsk)
2321: {
2322: return tsk->signal->nr_threads;
2323: }
2324:
2325: static inline bool thread_group_leader(struct task_struct *p)
2326: {
2327: return p->exit_signal >= 0;
2328: }
2329:
2330: /* Do to the insanities of de_thread it is possible for a process
2331: * to have the pid of the thread group leader without actually being
2332: * the thread group leader. For iteration through the pids in proc
2333: * all we care about is that we have a task with the appropriate
2334: * pid, we don't actually care if we have the right task.
2335: */
2336: static inline int has_group_leader_pid(struct task_struct *p)
2337: {
2338: return p->pid == p->tgid;
2339: }
2340:
2341: static inline
2342: int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2343: {
2344: return p1->tgid == p2->tgid;
2345: }
2346:
2347: static inline struct task_struct *next_thread(const struct task_struct *p)
2348: {
2349: return list_entry_rcu(p->thread_group.next,
2350: struct task_struct, thread_group);
2351: }
2352:
2353: static inline int thread_group_empty(struct task_struct *p)
2354: {
2355: return list_empty(&p->thread_group);
2356: }
2357:
2358: #define delay_group_leader(p) \
2359: (thread_group_leader(p) && !thread_group_empty(p))
2360:
2361: /*
2362: * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2363: * subscriptions and synchronises with wait4(). Also used in procfs. Also
2364: * pins the final release of task.io_context. Also protects ->cpuset and
2365: * ->cgroup.subsys[].
2366: *
2367: * Nests both inside and outside of read_lock(&tasklist_lock).
2368: * It must not be nested with write_lock_irq(&tasklist_lock),
2369: * neither inside nor outside.
2370: */
2371: static inline void task_lock(struct task_struct *p)
2372: {
2373: spin_lock(&p->alloc_lock);
2374: }
2375:
2376: static inline void task_unlock(struct task_struct *p)
2377: {
2378: spin_unlock(&p->alloc_lock);
2379: }
2380:
2381: extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2382: unsigned long *flags);
2383:
2384: #define lock_task_sighand(tsk, flags) \
2385: ({ struct sighand_struct *__ss; \
2386: __cond_lock(&(tsk)->sighand->siglock, \
2387: (__ss = __lock_task_sighand(tsk, flags))); \
2388: __ss; \
2389: }) \
2390:
2391: static inline void unlock_task_sighand(struct task_struct *tsk,
2392: unsigned long *flags)
2393: {
2394: spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2395: }
2396:
2397: /* See the declaration of threadgroup_fork_lock in signal_struct. */
2398: #ifdef CONFIG_CGROUPS
2399: static inline void threadgroup_fork_read_lock(struct task_struct *tsk)
2400: {
2401: down_read(&tsk->signal->threadgroup_fork_lock);
2402: }
2403: static inline void threadgroup_fork_read_unlock(struct task_struct *tsk)
2404: {
2405: up_read(&tsk->signal->threadgroup_fork_lock);
2406: }
2407: static inline void threadgroup_fork_write_lock(struct task_struct *tsk)
2408: {
2409: down_write(&tsk->signal->threadgroup_fork_lock);
2410: }
2411: static inline void threadgroup_fork_write_unlock(struct task_struct *tsk)
2412: {
2413: up_write(&tsk->signal->threadgroup_fork_lock);
2414: }
2415: #else
2416: static inline void threadgroup_fork_read_lock(struct task_struct *tsk) {}
2417: static inline void threadgroup_fork_read_unlock(struct task_struct *tsk) {}
2418: static inline void threadgroup_fork_write_lock(struct task_struct *tsk) {}
2419: static inline void threadgroup_fork_write_unlock(struct task_struct *tsk) {}
2420: #endif
2421:
2422: #ifndef __HAVE_THREAD_FUNCTIONS
2423:
2424: #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2425: #define task_stack_page(task) ((task)->stack)
2426:
2427: static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2428: {
2429: *task_thread_info(p) = *task_thread_info(org);
2430: task_thread_info(p)->task = p;
2431: }
2432:
2433: static inline unsigned long *end_of_stack(struct task_struct *p)
2434: {
2435: return (unsigned long *)(task_thread_info(p) + 1);
2436: }
2437:
2438: #endif
2439:
2440: static inline int object_is_on_stack(void *obj)
2441: {
2442: void *stack = task_stack_page(current);
2443:
2444: return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2445: }
2446:
2447: extern void thread_info_cache_init(void);
2448:
2449: #ifdef CONFIG_DEBUG_STACK_USAGE
2450: static inline unsigned long stack_not_used(struct task_struct *p)
2451: {
2452: unsigned long *n = end_of_stack(p);
2453:
2454: do { /* Skip over canary */
2455: n++;
2456: } while (!*n);
2457:
2458: return (unsigned long)n - (unsigned long)end_of_stack(p);
2459: }
2460: #endif
2461:
2462: /* set thread flags in other task's structures
2463: * - see asm/thread_info.h for TIF_xxxx flags available
2464: */
2465: static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2466: {
2467: set_ti_thread_flag(task_thread_info(tsk), flag);
2468: }
2469:
2470: static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2471: {
2472: clear_ti_thread_flag(task_thread_info(tsk), flag);
2473: }
2474:
2475: static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2476: {
2477: return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2478: }
2479:
2480: static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2481: {
2482: return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2483: }
2484:
2485: static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2486: {
2487: return test_ti_thread_flag(task_thread_info(tsk), flag);
2488: }
2489:
2490: static inline void set_tsk_need_resched(struct task_struct *tsk)
2491: {
2492: set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2493: }
2494:
2495: static inline void clear_tsk_need_resched(struct task_struct *tsk)
2496: {
2497: clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2498: }
2499:
2500: static inline int test_tsk_need_resched(struct task_struct *tsk)
2501: {
2502: return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2503: }
2504:
2505: static inline int restart_syscall(void)
2506: {
2507: set_tsk_thread_flag(current, TIF_SIGPENDING);
2508: return -ERESTARTNOINTR;
2509: }
2510:
2511: static inline int signal_pending(struct task_struct *p)
2512: {
2513: return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2514: }
2515:
2516: static inline int __fatal_signal_pending(struct task_struct *p)
2517: {
2518: return unlikely(sigismember(&p->pending.signal, SIGKILL));
2519: }
2520:
2521: static inline int fatal_signal_pending(struct task_struct *p)
2522: {
2523: return signal_pending(p) && __fatal_signal_pending(p);
2524: }
2525:
2526: static inline int signal_pending_state(long state, struct task_struct *p)
2527: {
2528: if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2529: return 0;
2530: if (!signal_pending(p))
2531: return 0;
2532:
2533: return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2534: }
2535:
2536: static inline int need_resched(void)
2537: {
2538: return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2539: }
2540:
2541: /*
2542: * cond_resched() and cond_resched_lock(): latency reduction via
2543: * explicit rescheduling in places that are safe. The return
2544: * value indicates whether a reschedule was done in fact.
2545: * cond_resched_lock() will drop the spinlock before scheduling,
2546: * cond_resched_softirq() will enable bhs before scheduling.
2547: */
2548: extern int _cond_resched(void);
2549:
2550: #define cond_resched() ({ \
2551: __might_sleep(__FILE__, __LINE__, 0); \
2552: _cond_resched(); \
2553: })
2554:
2555: extern int __cond_resched_lock(spinlock_t *lock);
2556:
2557: #ifdef CONFIG_PREEMPT_COUNT
2558: #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2559: #else
2560: #define PREEMPT_LOCK_OFFSET 0
2561: #endif
2562:
2563: #define cond_resched_lock(lock) ({ \
2564: __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2565: __cond_resched_lock(lock); \
2566: })
2567:
2568: extern int __cond_resched_softirq(void);
2569:
2570: #define cond_resched_softirq() ({ \
2571: __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2572: __cond_resched_softirq(); \
2573: })
2574:
2575: /*
2576: * Does a critical section need to be broken due to another
2577: * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2578: * but a general need for low latency)
2579: */
2580: static inline int spin_needbreak(spinlock_t *lock)
2581: {
2582: #ifdef CONFIG_PREEMPT
2583: return spin_is_contended(lock);
2584: #else
2585: return 0;
2586: #endif
2587: }
2588:
2589: /*
2590: * Thread group CPU time accounting.
2591: */
2592: void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2593: void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2594:
2595: static inline void thread_group_cputime_init(struct signal_struct *sig)
2596: {
2597: raw_spin_lock_init(&sig->cputimer.lock);
2598: }
2599:
2600: /*
2601: * Reevaluate whether the task has signals pending delivery.
2602: * Wake the task if so.
2603: * This is required every time the blocked sigset_t changes.
2604: * callers must hold sighand->siglock.
2605: */
2606: extern void recalc_sigpending_and_wake(struct task_struct *t);
2607: extern void recalc_sigpending(void);
2608:
2609: extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2610:
2611: static inline void signal_wake_up(struct task_struct *t, bool resume)
2612: {
2613: signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2614: }
2615: static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2616: {
2617: signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2618: }
2619:
2620: /*
2621: * Wrappers for p->thread_info->cpu access. No-op on UP.
2622: */
2623: #ifdef CONFIG_SMP
2624:
2625: static inline unsigned int task_cpu(const struct task_struct *p)
2626: {
2627: return task_thread_info(p)->cpu;
2628: }
2629:
2630: extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2631:
2632: #else
2633:
2634: static inline unsigned int task_cpu(const struct task_struct *p)
2635: {
2636: return 0;
2637: }
2638:
2639: static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2640: {
2641: }
2642:
2643: #endif /* CONFIG_SMP */
2644:
2645: extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2646: extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2647:
2648: extern void normalize_rt_tasks(void);
2649:
2650: #ifdef CONFIG_CGROUP_SCHED
2651:
2652: extern struct task_group root_task_group;
2653:
2654: extern struct task_group *sched_create_group(struct task_group *parent);
2655: extern void sched_destroy_group(struct task_group *tg);
2656: extern void sched_move_task(struct task_struct *tsk);
2657: #ifdef CONFIG_FAIR_GROUP_SCHED
2658: extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2659: extern unsigned long sched_group_shares(struct task_group *tg);
2660: #endif
2661: #ifdef CONFIG_RT_GROUP_SCHED
2662: extern int sched_group_set_rt_runtime(struct task_group *tg,
2663: long rt_runtime_us);
2664: extern long sched_group_rt_runtime(struct task_group *tg);
2665: extern int sched_group_set_rt_period(struct task_group *tg,
2666: long rt_period_us);
2667: extern long sched_group_rt_period(struct task_group *tg);
2668: extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2669: #endif
2670: #endif /* CONFIG_CGROUP_SCHED */
2671:
2672: extern int task_can_switch_user(struct user_struct *up,
2673: struct task_struct *tsk);
2674:
2675: #ifdef CONFIG_TASK_XACCT
2676: static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2677: {
2678: tsk->ioac.rchar += amt;
2679: }
2680:
2681: static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2682: {
2683: tsk->ioac.wchar += amt;
2684: }
2685:
2686: static inline void inc_syscr(struct task_struct *tsk)
2687: {
2688: tsk->ioac.syscr++;
2689: }
2690:
2691: static inline void inc_syscw(struct task_struct *tsk)
2692: {
2693: tsk->ioac.syscw++;
2694: }
2695: #else
2696: static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2697: {
2698: }
2699:
2700: static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2701: {
2702: }
2703:
2704: static inline void inc_syscr(struct task_struct *tsk)
2705: {
2706: }
2707:
2708: static inline void inc_syscw(struct task_struct *tsk)
2709: {
2710: }
2711: #endif
2712:
2713: #ifndef TASK_SIZE_OF
2714: #define TASK_SIZE_OF(tsk) TASK_SIZE
2715: #endif
2716:
2717: #ifdef CONFIG_MM_OWNER
2718: extern void mm_update_next_owner(struct mm_struct *mm);
2719: extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2720: #else
2721: static inline void mm_update_next_owner(struct mm_struct *mm)
2722: {
2723: }
2724:
2725: static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2726: {
2727: }
2728: #endif /* CONFIG_MM_OWNER */
2729:
2730: static inline unsigned long task_rlimit(const struct task_struct *tsk,
2731: unsigned int limit)
2732: {
2733: return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2734: }
2735:
2736: static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2737: unsigned int limit)
2738: {
2739: return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2740: }
2741:
2742: static inline unsigned long rlimit(unsigned int limit)
2743: {
2744: return task_rlimit(current, limit);
2745: }
2746:
2747: static inline unsigned long rlimit_max(unsigned int limit)
2748: {
2749: return task_rlimit_max(current, limit);
2750: }
2751:
2752: #endif /* __KERNEL__ */
2753:
2754: #endif